Claude Fable 5.2 lands: beats Opus 5.5 and GPT-6 across agent benchmarks
accourtt · reddit · 2026-11-10
Claude Fable 5.2 is out, and its leaked benchmark table shows broad gains over Fable 5.1, Opus 5.5, GPT-6 Astra, and Gemini 4 Carbon:
- Agentic coding: 49.1% on FrontierCode 2.0, 69.9% on FrontierSWE v2 — big jumps from 32.4%/56.3%.
- Science & security: 74.6% Terminal-Bench-Science, 76.3% CyScenarioBench, 89.6% CVE Rediscovery, 85.1% de novo protein binder design.
- Agents: 82.1% WANDR, 89.3% OSWorld 2.1 computer use, 64.9% RoboDojo Sim robotics (vs 27.6% prior).
- Knowledge work: 1785 Elo on AA-Briefcase, Remote Labor Index 41.3% vs 17.9% prior, 74.2% HLE with tools.
- One exception: AutomationBench business workflows at 51.4%, still behind Gemini 4 Carbon's 53.9%.
Full methodology per the Fable 5.2 and Mythos 5.2 system cards.
More from Models
- Early Tester: Fable 5.2 Built a 3,100-City Water Map in One Go, but Still Makes Silly Errors — alliekmiller · 2026-11-10
- Fable 5.2 burns so many tokens its 20% price cut is effectively a wash — addyosmani · 2026-11-10
- Opus formalizes CPython in 613K lines of Lean — and finds real Python bugs — pbaylies · 2026-11-10
- CursorBench 4.0: Fable 5.2 High effort hits 56.9% at $9.70 per task — SEO · 2026-11-10
- Rule of thumb: use Opus 5.5 for tasks you know, Fable 5.2 for creative exploration — EricBuess · 2026-11-10
- Claude Fable 5.2 tops FrontierSWE at 69.9%, beating GPT-6 Astra and Opus 5.5 at 1/6 the cost — aryaman2020 · 2026-11-10